Xyratex ClusterStor 6000 for HPC Storage Competing With DDN

Xyratex Ltd announced the ClusterStor 6000, an HPC fast storage solution delivering two times the Lustre file system performance over competing designs.

With an enterprise level resiliency architecture, ClusterStor 6000 provides high level in integrated solution reliability, availability and serviceability. It will be available exclusively through OEMs in the third quarter of this year.

It delivers a new standard in file system performance, scalability and efficiency. Its architecture brings together what were previously discrete server, network and storage platforms with their own separate software layers, into a single integrated modular scale out storage building block. This approach enables the system to scale both performance and capacity while reducing space, power and administrative overhead.

ClusterStor 6000 Scalable Storage Unit (SSU) is using building block that enables rapid deployment, scale and productivity. The SSU is only five rack units high and delivers up to six gigabytes/sec of sustained read and write file system performance. It offers support for installations with linear performance scalability from 6 gigabytes per second to one terabyte per second file system processing capability, as well as linear data storage capacity from terabytes to tens of petabytes.

"The ClusterStor product family of scale-out HPC data storage solutions features a tightly integrated distributed computing architecture. Xyratex has integrated Lustre into storage appliances designed to operate as turnkey solutions, contrasted with the traditional piecemeal, less-productive HPC storage solutions. The integration of heterogeneous traditional nodes into high performance, dense platforms results in a reduced footprint and the foundation for lower cooling and infrastructure costs," said Steve Conway, research VP, High Performance Computing, IDC. "Xyratex is involved in some of the largest and most challenging storage deployments in the world, at least one of which aims to become the first high-density platform to scale to 1TB/second sustained file system performance."

ClusterStor 6000 is designed for users who want to build petascale computing solutions for HPC applications in areas such as scientific research, simulation, climate modeling, and energy exploration.

"We’ve been using ClusterStor 3000 as part of our GENCI/PRACE funded (GENCI/PRACE funded ‘Exascale I/O prototype’ is funded in part by the EUs 7th Framework Programme (FP7/2007-2013) under grant agreement no. FP7-RI-261557) ‘Exascale I/O prototype’ for the last 3-6 months and are pleased with the Lustre performance numbers being achieved as well as the ease of installation and management," said Jacques-Charles Lafoucriere, chief of services, CEA. "Our next phase of the project is a simple upgrade to the ClusterStor 6000 OSS with the next generation Intel E5 processors where we fully expect a significant increase in Lustre system level improvements."

ClusterStor 6000 deployment is a snap – it only requires physical installation and simple setup steps. Instead of spending weeks gathering separate hardware and software, resolving interoperability issues and testing to get a Lustre environment into production, users can be productive with ClusterStor in hours. It includes pre-installation of all internal software components, configuration of the integrated system, system reliability testing and ships ready to go to work.

ClusterStor Manager is a system management application that is part of ClusterStor’s distributed management framework and is responsible for pulling everything together as a singly managed system. It consolidates management of the entire storage infrastructure, RAID data protection layer, operating system, and the Lustre file system into a single, easy-to-use, administrator interface. From initial installation, creation of file systems, ongoing monitoring and management, to expansion and software upgrades, ClusterStor Manager brings the entire storage cluster solution to a single pane of glass. This provides system visibility to the Lustre system administrator.

"The ClusterStor 6000 is a truly next generation storage system developed to address the pervasive need for massively scalable storage systems within the HPC and broader Big Data market," said Ken Claffey, Xyratex senior director of the ClusterStor Business Line. "ClusterStor provides two levels of integration. The first combines file systems and RAID data protection onto our embedded server modules and the second consolidates these capabilities into our high density storage platform as a Scalable Storage Unit. These levels of integration enable linear scalability delivering 1TB/sec file system performance with 30% less power, and floor space of competitive systems. With the performance needed in many new applications, the legacy approach to HPC data storage doesn’t work."